I modified the config.h by changing all the ifort to mpif90, and the
compiling is successful. I will see if it runs ok.
On Thu, Feb 11, 2010 at 7:10 PM, Jason Swails <jason.swails.gmail.com>wrote:
> Hello,
>
> The configure files are probably out of date for the current versions
> of the compilers, so you may have to modify the config.h file created
> by the configure script.
>
> On Thu, Feb 11, 2010 at 6:56 PM, Dian Jiao <oscarjiao.gmail.com> wrote:
> > Hi,
> >
> > I am having trouble compiling PMEMD (v9) on a cluster.
> >
> > After loading the compiler module (intel) and mpi module (mvapich2), I
> did
> > "make" to compile it.
> >
> > When I used ifort 9.1, the error msg was:
> >
> > error: IPO link: can not find -lmtl_common.
>
> If you get errors like this, simply remove that flag from the config.h
> file wherever it occurs and retry (no need to make clean after these
> changes, the compile will pick up where it left off), though have a
> look at my suggestions below.
>
> >
> > However if I change it to ifort 10.1, I got a different error:
> >
> > libmpich.so undefined reference to aio_read64/aio_error64/aio_write64..
>
> Your cluster should have appropriate compiler wrappers that will
> correctly link the mpi libraries, specifically mpif90 (or mpif77) for
> Fortran 90 code. If you haven't already, change all instances of
> "ifort" to "mpif90" in the config.h file. One thing you'll want to
> make sure of, though, is that the mpif90 you're using in config.h
> wraps the proper compiler you're trying to use (the command "mpif90
> -show" will display the invoked compiler with all linked libraries).
> This can become slightly messy if the system you're trying to build on
> has a jumbled mess of different MPI/compiler combinations, however, so
> some caution should be taken.
>
> >
> > I did on some other supercomputers before, it never failed.
> >
> > Has anyone seen this before?
> >
> > Thanks in advance.
> >
> > Dian
> > _______________________________________________
> > AMBER mailing list
> > AMBER.ambermd.org
> > http://lists.ambermd.org/mailman/listinfo/amber
> >
>
> Good luck!
> Jason
>
> --
> ---------------------------------------
> Jason M. Swails
> Quantum Theory Project,
> University of Florida
> Ph.D. Graduate Student
> 352-392-4032
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Feb 11 2010 - 19:30:02 PST
